AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Visual Transformers: Where Do Transformers Really Belong in Vision Models?

Bichen Wu, Chenfeng Xu, Xiaoliang Dai, Alvin Wan, Peizhao Zhang, Zhicheng Yan, Masayoshi Tomizuka, Joseph E. Gonzalez, Kurt Keutzer, Peter Vajda

ICCV 2021 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 112 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use filter-based tokenizer instead of pooling-based (ablation)VT-ResNet-18 · ImageNet val (VT-ResNet-18)top-1 accuracy (%): 70.6 -> 72.0Improved
use filter-based tokenizer instead of pooling-based (ablation)VT-ResNet-34 · ImageNet val (VT-ResNet-34)top-1 accuracy (%): 73.5 -> 74.8Improved
filter-based vs clustering-based tokenizer (ablation)VT-ResNet · ImageNet valtop-1 accuracy (%): R18 71.5 -> 72.0; R34 75.1 -> 74.8Mixed
add recurrent tokenizers (ablation)VT-ResNet-18 · ImageNet val (VT-ResNet-18)top-1 accuracy (%): 71.4 -> 72.0Improved
add recurrent tokenizers (ablation)VT-ResNet-34 · ImageNet val (VT-ResNet-34)top-1 accuracy (%): 74.5 -> 74.8Improved
use transformer instead of no token interaction (ablation)VT-ResNet-18 · ImageNet val (VT-ResNet-18)top-1 accuracy (%): 68.8 -> 72.0Improved
use transformer instead of graph convolution (ablation)VT-ResNet-34 · ImageNet val (VT-ResNet-34)top-1 accuracy (%): 73.6 -> 74.8Improved
increase visual tokens from 16 to 64 (ablation)VT-ResNet-18 · ImageNet val (VT-ResNet-18)top-1 accuracy (%): 72.0 -> 72.0No clear effect
increase visual tokens from 16 to 32 (ablation)VT-ResNet-34 · ImageNet val (VT-ResNet-34)top-1 accuracy (%): 74.8 -> 75.1No clear effect
add projector back to feature map (ablation)VT-ResNet-18 · ImageNet val (VT-ResNet-18)top-1 accuracy (%): 70.9 -> 72.0Improved
train with advanced recipe instead of defaultVT-ResNet-18 · ImageNet valtop-1 accuracy (%): 72.0 -> 76.8Improved
replace FPN with VT-FPNResNet-50 FPN · COCO-stuff valmIoU (%): 40.78 -> 41.00Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy